Abstract

Graphical Abstract

Open in a separate window

1. Introduction

The extracellular matrix (ECM) is one of the key elements within a solid tumor that restricts drug penetration into tumor cells. This aspect of tumor biology was recognized for decades and has been extensively reviewed^1–3 as have efforts to remove the ECM with enzymes or collagen/ hyaluronan synthesis inhibitors.^4–10 Movement of macromolecular oncotherapies, such as antibodies and nanoparticles, are more sensitive to the ECM than low molecular weight drugs so efforts to modify the ECM will be most beneficial to the therapeutic activity of these agents.

Several tumors including breast, pancreatic, colorectal, ovarian, and lung exhibit a dense ECM where higher collagen or hyaluronan content correlates with poor prognosis.^11,12 Cancer associated fibroblasts produce large amounts of ECM components which are associated with tumor promoting activities including: angiogenesis^13,14, cell proliferation⁴, and the creation of ‘highways’ for extravasation of metastatic cells or intravasation of pro-tumor immune cells (Figure 1).^4,11 Tumor ECM, which is composed primarily of collagen and hyaluronic acid (HA), contributes to several aspects of the tumor microenvironment that obstructs drug delivery, including: 1. Compression of blood vessels, decreasing blood perfusion and restricting adequate vascular access to portions of the tumor;¹⁵ 2. Limiting drug trafficking through the tumor matrix interstitium preventing drugs from reaching their cellular targets;^16,17 and 3. Maintenance of a high interstitial fluid pressure (IFP) generated by plasma leakage from blood, mechanical stress within the tumor and a lack of proper lymphatic drainage.¹⁸ The proliferation of tumor cells within the confined microenvironment further contributes to the mechanical stress within the tumor microenvironment.¹⁸ All of these restrict the movement of drugs from the blood into tumor cells (Figure 2).^15,17

Open in a separate window

Figure 1:

Elevated ECM enables tumor promoting activities.

ECM in tumors causes: 1. Increased angiogenesis to supply nutrients for the growing tumor; 2. Growth factor signaling leading to cell proliferation; 3. Recruitment of anti-inflammatory, protumor immune cells; and 4. Creating of a path for metastatic cells to extravasate from the primary tumor.

Open in a separate window

Figure 2:

A dense tumor extracellular matrix limits adequate delivery of drugs.

Blood vessels can become compressed preventing drugs from reaching the tumor. Those drugs that are able to enter the tumor microenvironment are faced with a high interstitial fluid pressure and limited interstitial drug transport.

In theory, approaches to decrease the tumor ECM should increase drug penetration and enable higher drug concentrations within tumor cells. There is a subtle aspect to this approach. Complete removal of all biopolymers in the tumor interstitial space might result in collapse of the tumor and a decrease in drug penetration; as such the concept of normalization of the ECM has been suggested to be the goal of such therapies (3,4,10,16,17,18). In this review we highlight key approaches for ECM degradation but focus on the activity of collagenases since these enzymes are less widely used yet can be mechanistically distinguished from the most advanced form of matrix degradation achieved through the reduction of hyaluronic acid (HA).

2. HA Modification

One of the earliest strategies to deplete the tumor ECM is to target hyaluronic acid (HA) either by preventing its production or by digesting what is present in the tumor ECM.¹⁹ The utility of HA inhibition has been comprehensively reviewed elsewhere.^6,8 Notably, 4-methylumbelliferone (4-MU) has been utilized to prevent HA formation in a variety of tumors by depleting the substrate required for hyaluronan synthesis.^20–22 It also downregulates expression of hyaluronan synthase 2 and 3.²¹ The use of 4-MU illustrates the potential dual effect achieved from inhibiting tumor ECM. It directly decreased tumor growth and metastasis in prostate, liver, breast and skin cancer due to a reduction in tumor HA.⁶ In addition, treatment of tumored animals with 4-MU enhanced the extravasation of nanoparticles from the blood vessels into the tumor. For example, a combination therapy of a liposome-encapsulated 4-MU prodrug with liposomal doxorubicin decreased tumor volume and increased overall survival compared to liposomal doxorubicin alone in an orthotopic murine tumor.²³ The therapeutic enhancement was a result of the improved distribution of liposomal doxorubicin due to a reduction in tumor ECM.²³ Oral 4-MU (Cantabiline®), is approved in Europe and Asia as a dietary supplement to increase bile flow. It is a safe drug but has very low oral bioavailability (<3%)⁸ and its efficacy in oncology remains an open question.

Alternatively, hyaluronidase has been extensively used to enzymatically digest HA.²⁴ Hyaluronidase efficiently degrades large HA polymers comprised of up to 25,000 monosaccharide units down to simple polysaccharides under 10 units.²⁵ Hyaluronidase was explored in oncology clinical trials starting in the 1980s,²⁶ where pretreatment displayed improved outcomes in head and neck, brain, bladder, and gastrointestinal cancer clinical trials.³ Notably, 1 in 3 patients developed immune reactions to the bovine hyaluronidase.²⁶ Detailed assessments of the utility of hyaluronidase for improving interstitial and tumor penetration were previously written. ^{3,5,7,9,10,27} A pegylated human hyaluronidase (PEGPH20) introduced by Halozyme Therapeutics has advanced to late stage clinical trials. This polymer-modified formulation of hyaluronidase reduces immune recognition and promotes extended circulation, giving the enzyme ample time to accumulate in the tumor ECM.²⁷ PEGPH20 in combination with gemcitabine and nab-paclitaxel (Abraxane®) in a phase II metastatic pancreatic cancer trial showed an increased overall survival in treated patients compared to the control group.²⁸ This is currently the most advanced ECM removal strategy. It will be interesting to learn if the beneficial effects are replicated in the Phase III trial, or in an upcoming gastric cancer trial combining PEGPH20 with Anti-PDL1 immunotherapy.²⁹

3. Collagen Modification

Similar to efforts to remove HA from the tumor matrix, attempts have also been made to limit the synthesis or promote the breakdown of extracellular tumor collagen. Reducing collagen synthesis has been achieved most aggressively through TGF-β inhibition, altering its regulatory role in collagen synthesis.³⁰ The Jain group has spearheaded this through the use of Losartan, an angiotensin II inhibitor, and the use of anti-TGF-β antibodies. Losartan mediates angiotensin II type receptor downregulation of TGF-β activators such as thrombospondin-1 causing a reduction in TGF-β signaling.³⁰ TGF-β is an actively pursued target in oncology so the anti-tumor effects from its inhibition may work in tandem with the benefits of reducing tumor collagen levels. TGF-β inhibition with Losartan led to an increase in drug penetration and subsequent improvement in survival in multiple murine tumors. ^30–32 Lorsartan also targets hyaluronan and cancer associated fibroblasts.^30,31 Nonetheless, the preclinical success was mirrored in a small phase II clinical trial in pancreatic cancer testing the benefits of Losartan in combination with the FOLFIRINOX (leucovorin, 5-fluorouracil, irinotecan, oxaliplatin) chemotherapy combination where over 50% of eligible patients identified by radiographic imaging were able to have their tumors resected leaving a margin that had no detectable tumor cells.^18,33

Another mode to limit collagen deposition is through inhibition of lysyl oxidase-like-2 (LOXL2) activity. LOXL2 crosslinks collagen to the fibrillar form found in the ECM and an antibody against LOXL2 reduced the number of collagen crosslinks and overall tumor burden in murine xenographs.³⁴ LOXL2 inhibition also caused a decrease in activated fibroblasts and production of growth factors and cytokines involved in cell signaling.³⁴ Thus, as with TGF-β blockade, the anti-tumor effects of LOXL2 inhibition occur in tandem to the benefits of collagen reduction. Despite the preclinical success there is concern that this approach of blocking matrix synthesis may be therapeutically limited to early tumor stages prior to the establishment of the characteristically dense matrix. For instance, a Phase II trial in metastatic pancreatic cancer patients that combined simtuzumab (anti-LOXL2) with gemcitabine failed to improve clinical outcomes likely due to the advanced stage of the cancer.³⁵ Several other antifibrotic drugs have been investigated to reduce tumor ECM including tranilast,³⁶ pirfenidone,³⁷ fasudil,³⁸ and metformin.³⁹ The expanded Losartan trial will be pivotal to understand if inhibition of ECM by repurposing low molecular weight, orally available molecules can translate into the clinic.

The alternative approach to deplete tumor collagen is through the use of collagenases. Collagenases were explored in the 1980s to dissolve excess collagen in patients with severe back pain via collagenase injection into spinal discs.⁴⁰ The clinical utility of collagenases is validated by the approval of Xiaflex®, a bacterial Clostridium Histolyticum collagenase, for the treatment of Dupuytren’s contracture as an injection against thickening of collagen tissue within the hand. However, the clinical utility of collagenases in improving cancer therapy is much less well established despite collagen being the most abundant tumor ECM component.

In theory, the use of collagenases would be especially attractive for improving drug penetration but unlike hyaluronidase which degrades linear hyaluronan down to short oligosaccharides, collagenases cleave at distinct sites along the collagen triple helix^37–40 leaving behind large subunits of approximately 10–95 kDa (Figure 3).^41,42 These cleavage products may not effectively separate from collagen fibers resulting in only microscopic local changes to the collagen structure with collagenase therapy.⁴³ Preclinical work to evaluate the use of collagenases in oncology, however, indicated that collagenase treatment improved drug diffusion and penetration in treated tumors (Table 1-Table 6). The challenge for interpreting or comparing results from the studies reviewed in the following sections is that they varied in enzyme exposure (amount and time), route of injection, tumor model, penetration criteria, and type of drug. This makes it difficult to predict the potential clinical utility of injected collagenases in cancer therapy or to evaluate the promise and limitations of the approach.

Open in a separate window

Figure 3:

Cleavage of extracellular collagen and hyaluronan.

Extracellular hyaluronan and collagen are cleaved by hyaluronidase and collagenase, respectively. Hyaluronidase cleaves at hexosaminidic bonds between β-(1,3)-D-glucuronic acid and β-(1,4)-N-acteyl-D-glycosamine, producing easily cleared oligosaccharide fragments. Bacterial and human collagenases (matrix metalloproteinases-1,3, 8) cleave at defined sites along the collagen triple helix.⁴² Large collagen cleavage fragments may not effectively be removed from collagen fibers.

4. Diffusion as predictor of drug accessibility in tumors

Diffusion is a key factor that controls delivery of therapeutics into the tumor core. The high IFP and torturous ECM within tumors limits the ability of drugs to freely diffuse.⁴⁴ Furthermore, the amount of collagen inversely correlates with tumor diffusivity.¹⁷ Studies examining the effects of collagenase on drug diffusion focus primarily on protein therapeutics. The findings are especially relevant because protein therapeutics are the most successful class of clinically approved macromolecules in recent years⁴⁵ and will require improved penetration for the continued clinical success.

Collagenase treatment increased the diffusion of macromolecules in tumors by a modest 2-fold in a majority of studies (Table 1). Alexandrakis et al.⁴⁶ observed that molecules moving through a Mu89 melanoma tumor undergo both a rapid and slow diffusion. The slower diffusion occurs in parts of the tumor with heightened ECM deposition. Intratumoral (IT) collagenase treatment removed this impediment and caused non-specific IgG antibodies to shift from slow to more rapid diffusion by 1.3-fold within a tumor. Netti et al.⁴⁷ also investigated the role of ECM components on transport in four different tumors. HST sarcoma and U87 glioblastoma tumors had 2 to 5-fold more collagen relative to tumor mass than the LS174T and MCalV tumors investigated. Higher collagen content in the HST and U87 tumors corresponded with a 2-fold decrease in diffusivity of an IgG antibody. In these tumors, intratumorally injected collagenase doubled the diffusion coefficient for a labeled IgG. Notably, despite the improvement from collagenase treatment, it only restored transport to levels seen in LS174T and MCalV tumors and a larger effect may be required to significantly improve delivery to therapeutically relevant levels.

Eikenes and coworkers⁴⁸ used an intravenous (IV) injection of collagenase but also obtained a comparable 2-fold increase in diffusion for a 150 kDa FITC-Dextran. Their studies were completed using an osteosarcoma (OHS) tumor model. The change in diffusion following collagenase treatment persisted for at least 48 hours. This is promising because it would allow time for drugs to traffic into tumors and exploit the increased diffusivity to penetrate into the tumor core.

Magzoub et al. is the only study to obtain an increase in diffusion greater that 2-fold.⁴⁹ They reported a 10-fold increase in diffusion for a 500 kDa dextran at a depth of 2 mm into a tumor. While previous work relied on the more conventional dorsal skinfold window chamber tumor model coupled with fluorescence resonance after photobleaching (FRAP) to determine diffusivity (Table 1), Magzoub designed a fiberoptic probe to intravitally measure diffusion using a more representative melanoma tumor model. Their innovation enabled photobleaching measurements for diffusion at multiple depths within a tumor. Larger macromolecules do not penetrate deeply into dense tissues thus even a minor absolute change in the penetration could lead to a large relative change in penetration, which may be why the magnitude of this scale has not been replicated elsewhere. This is underscored by the fact that at a more superficial depth or with a smaller particle the magnitude is reduced to what was observed in previous studies. At a depth of 0.5 mm, along the tumor periphery, the increase in diffusion for the 500 kDa dextran was only 2-fold. As well, bovine serum albumin whose mass is an order of magnitude less, had an increase in diffusion of only 2-fold at all depths studied. Enhancements in tumor diffusion are overall promising but do not directly address if a 2-fold increase in diffusivity parallels an increase in total drug within tumors.

5. Drug Uptake

6. Spheroids

Measuring the effects of collagenase treatment in vivo can be experimentally challenging, timely and costly. The common dorsal skinfold window chamber tumor model, for example, requires complex surgery and specialized equipment to be used effectively.^73,74 There is a need for suitable in vitro systems to assess the benefits of collagenases prior to moving in vivo. Traditional 2-D cell culture lacks a suitable architecture and ECM to reliably study the effects of matrix breakdown.^75,76 3-D tumor spheroids are the preferred in vitro model for collagenase therapy since they mimic several aspects of the avascular portions of a tumor including epithelial tight junctions, an inhibitory ECM, cellular heterogeneity and a proliferating and quiescent region along with a necrotic core.^77–79 The use of tumor spheroids is contentious as a means to evaluate drug delivery systems, however drug penetration studies in tumor spheroids have demonstrated a 2 to 11-fold increase in particle delivery following collagenase treatment which is on the same order as what is observed in vivo (Table 5). However, as with in vivo experiments, spheroid studies seldom report quantifiable changes to tumor collagen with collagenase treatment.

Eikenes et al.⁴⁸ found that despite a 2-fold increase in diffusion of a 150 kDa dextran in vivo, there was no significant change in dextran diffusion in vitro in tumor spheroids. However, Elkenes observed the larger 2 MDa dextran showed an almost 2-fold increase in diffusion following collagenase treatment in the same OHS spheroids. This suggests that spheroids may be useful predictors for the diffusion of larger particles but the diffusion of lower molecular weight drugs may not be sufficiently inhibited by the spheroid ECM to observe the effects of collagenase treatment.

Goodman et al.⁸⁰ and Cui et al.⁸¹ both measured penetration of various nanoparticles as a function of their diameter and observed that collagenase treatment increased penetration in a size-dependent manner. Goodman et al. found a 7, 12, 3, and 1.5-fold increase in fluorescence in the spheroid core for 20, 40, 100 and 200 nm fluorescently labeled polystyrene nanoparticles respectively. The effect of collagenase treatment on particle uptake peaks for 40 nm particles and is much less pronounced for the 100 and 200 nm particles. When collagenase was attached to the surface of the 100 nm particle there was a 4-fold further increase (totaling ~12 fold) in fluorescence signal beyond the spheroid periphery compared to an equivalent particle given with free collagenase.

Cui et al.⁸¹ coated albumin nanoparticles with collagenase and found a 6% and 27% increase in penetration for 100 and 200 nm particles respectively. Spheroids were imaged after a 96-hour exposure to collagenase-coated nanoparticles to determine the localization of the nanoparticles. It is important to note that the 200 nm particles exhibited approximately 2-fold greater collagenase activity than the 100 nm particles which could account for the larger percent increase in localization. The authors did not quantify the total fluorescent signal in spheroids which makes it difficult to compare the magnitude of change to what was seen by the nanoparticles in Goodman et al.. However, the authors noted that only the 100 nm collagenase-coated nanoparticles displayed increased signal in the spheroid core (unquantified) which supports the importance of particle size for deep penetration within spheroids. These studies replicate the enhancement in nanoparticle penetration observed in vivo in Murty et al.⁶⁶ (Table 3) when collagenase is covalently attached to the nanoparticle.

Spheroids were also evaluated as models for determining enhancements in gene therapy following ECM clearance. In 9L rat glioma tumor spheroids, collagenase therapy showed no improvement in delivery of an AAVP tumor-targeting phage carrying a luciferase gene.⁸² However, the combination of collagenase and hyaluronidase increased luciferase activity almost 3-fold in these tumor spheroids. The magnitude is smaller than what was observed in vivo with luciferase and eGFP plasmids⁶⁹ but aligns with the finding that digestion of multiple ECM components can provide greater improvement in macromolecular delivery (Table 4).

Although the improvements in drug penetration with spheroids are consistent with those in vivo, their use is not without caveats. The relationship between spheroid collagen content and drug uptake is unclear. Spheroid morphology and response to external stimuli are extremely sensitive to the method used to produce the 3-D cells.⁷⁸ In addition, spheroid penetration studies are limited to tumor cells which can form spheroids at appropriate sizes to adequately restrict drugs. Spheroids have also not demonstrated the ability to measure changes in penetration of small molecule and protein therapeutics following ECM breakdown. More quantitative studies are needed using standardized methods for growth and viability, and that show that outcomes in spheroids replicate in vivo before spheroids can be considered fully vetted tools for assessing drug penetration.

7.0. Overall outlook; needs and benefits

It is undetermined if the moderate improvements in drug penetration observed in vivo with collagenase treatment in animals will translate to patients. There are major concerns regarding the toxicity of injected collagenase due to the potential for increased tumor metastasis,^83–91 degradation of collagen in healthy tissues,^62,92 and immune reactions against bacterial collagenases.⁶⁶ For instance, collagenases that could improve drug delivery when administered at 100–300 µg/mouse resulted in fatalities when injected at greater than 500 µg/mouse.^51,63 It is debatable whether the potential efficacy is enough to overcome such a narrow therapeutic window.

8. Ongoing needs to validate matrix reduction therapies to improve cancer treatment

The current literature indicates that collagenases can be both effective and tolerated at low doses in mice but to solidify these findings we must: 1. Better understand the changes to tumor matrix subsequent to collagenase treatment, particularly the clearance of collagen fragments, the rebound of collagen fibers to pre-treatment levels, and the functionality of tumor blood vessels; 2. Use more relevant tumor models which broaden the solid tumor types that are treated with collagenases; and 3. Move beyond unspecific bacterial collagenases. First, few studies quantified changes to tumor collagen content or organization following collagenase treatment. Those that did found little to no effect in collagen that correlates with changes in probe penetration (Table 1–4). Small (12 kDa) degradation fragments of collagen can separate from fibers and be directly measured in plasma whereas larger fragments require additional proteolytic cleavage.^93,96 The dose of collagenase used to improve drug infiltration may not allow adequate cleavage of collagen to liberate larger fragments.

For instance, one study validating the use of SHG to measure tumor collagen showed that an IT collagenase dose of 10 mg,⁹⁷ which is substantially greater than the doses used in a majority of studies measuring drug penetration, was required to observe dramatic changes in tumor SHG intensity. Lower collagenase doses may only exhibit minor changes to tumor collagen. This makes it difficult to know whether a change in total collagen or a change in collagen structure is needed or for how long that change must persist in order to observe increased probe penetration. In addition, there is little understanding of how collagenase treatment reduces tumor mechanical stress and normalizes drug perfusion. Changes in tumor IFP offer a surrogate for the robustness of the effect of collagenase treatment in some instances but a more complete understanding of what happens to collagen, the ECM architecture and tumor pressure after treatment is needed.

Second, the dorsal skinfold window chamber tumor model is the most widely used because it allows for intravital imaging. The problem is that this model is limited to tumors which can grow in that environment and for those which can, they may not exhibit a canonical array of tumor behaviors in that atypical microenvironment.⁷³ As such, traditional orthotropic and spontaneous tumors would serve as better predictors of the benefits of collagenase treatment. In these models, it is essential to demonstrate that an increase in particle penetration causes an increase in overall survival. The few studies that seek to address this show decreases in tumor volume but do not unequivocally show improved in vivo survival. Performing IV over IT injections of collagenase should also be prioritized since the need for defined tumor boundaries for a successful IT injection may limit the tumors that are candidates for matrix reduction therapies.

Third, studies discussed here use bacterial collagenase which exhibits activity on a variety of collagens.^98,99 Collagen I is the most abundant in vertebrates but there are 28 different types of collagen identified.¹¹ Several types of collagen including collagen I, II, III, IV, V, IX are implicated in cancer progression.^11,12 Therefore, it could be advantageous to tailor the collagenase to the tumor or create a mixture of collagenases that would digest collagens in many different tumor types. In this regard, matrix metalloproteinases (MMP), in particular, should be further explored for ECM degradation because they would limit the immunogenicity from a bacterial collagenase as well as offer a library of proteinases to better match the collagenase to the type of tumor collagen.

10. Conclusion

The increased exploration using macromolecules, nanoparticles, and viruses to treat cancer motivates the need for ways to improve penetration of these agents into tumors. The data for injected collagenase treatment suggest it to be relatively safe and modestly efficacious (Table 1–Table 6). However, before that potential can be fully realized, investigators will have to explicate how collagenase treatment affects the amount and structure of tumor collagen and how changes in collagen relate to alterations in the tumor microenvironment. This information could then be used to understand precisely what changes to the ECM are needed to enhance drug delivery as well as enable an appropriate selection of enzyme combinations and matrix metalloproteinases subtypes to maximize drug uptake in tumors. Making these rational improvements could digest a path forward for matrix reduction therapy as a clinical modality to enhance the delivery of drugs, macromolecules, and nanoparticles into solid tumors.

Acknowledgement:

References